Re: [nfsv4] More layout stuff: layout return on close
Benny Halevy <bhalevy@panasas.com> Wed, 01 December 2010 15:48 UTC
Return-Path: <bhalevy@panasas.com>
X-Original-To: nfsv4@core3.amsl.com
Delivered-To: nfsv4@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id C4F6C3A6C6E for <nfsv4@core3.amsl.com>; Wed, 1 Dec 2010 07:48:44 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.896
X-Spam-Level:
X-Spam-Status: No, score=-2.896 tagged_above=-999 required=5 tests=[AWL=-0.631, BAYES_00=-2.599, IP_NOT_FRIENDLY=0.334]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ryZrevtGOY3D for <nfsv4@core3.amsl.com>; Wed, 1 Dec 2010 07:48:43 -0800 (PST)
Received: from daytona.panasas.com (daytona.panasas.com [67.152.220.89]) by core3.amsl.com (Postfix) with ESMTP id BE9E73A6BBD for <nfsv4@ietf.org>; Wed, 1 Dec 2010 07:48:42 -0800 (PST)
Received: from lt.bhalevy.com ([172.17.33.19]) by daytona.panasas.com with Microsoft SMTPSVC(6.0.3790.4675); Wed, 1 Dec 2010 10:49:55 -0500
Message-ID: <4CF66EA2.7010407@panasas.com>
Date: Wed, 01 Dec 2010 17:49:54 +0200
From: Benny Halevy <bhalevy@panasas.com>
User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.12) Gecko/20101027 Fedora/3.1.6-1.fc13 Thunderbird/3.1.6
MIME-Version: 1.0
To: Trond Myklebust <Trond.Myklebust@netapp.com>
References: <1290552459.30508.4.camel@heimdal.trondhjem.org> <4CECC382.5020207@panasas.com> <4CECE104.5080009@desy.de> <1290610592.3234.20.camel@heimdal.trondhjem.org><4CEE283D.5040903@desy.de> <1290692948.2911.7.camel@heimdal.trondhjem.org> <4CEE8AA9.1030109@desy.de> <BF3BB6D12298F54B89C8DCC1E4073D8002A5F2B6@CORPUSMX50A.corp.emc.com> <4CF25270.5050609@panasas.com> <1290968847.12027.6.camel@heimdal.trondhjem.org> <BF3BB6D12298F54B89C8DCC1E4073D8002A5F323@CORPUSMX50A.corp.emc.com> <1291045387.12784.7.camel@heimdal.trondhjem.org> <BF3BB6D12298F54B89C8DCC1E4073D8002A5F42F@CORPUSMX50A.corp.emc.com> <1291053460.12784.14.camel@heimdal.trondhjem.org> <BF3BB6D12298F54B89C8DCC1E4073D8002A5F455@CORPUSMX50A.corp.emc.com> <1291056819.12784.30.camel@heimdal.trondhjem.org> <4CF523D8.4040206@panasas.com> <1291155293.2998.33.camel@heimdal.trondhjem.org>
In-Reply-To: <1291155293.2998.33.camel@heimdal.trondhjem.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-OriginalArrivalTime: 01 Dec 2010 15:49:55.0862 (UTC) FILETIME=[6694DB60:01CB916F]
Cc: nfsv4@ietf.org
Subject: Re: [nfsv4] More layout stuff: layout return on close
X-BeenThere: nfsv4@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: NFSv4 Working Group <nfsv4.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/nfsv4>, <mailto:nfsv4-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/nfsv4>
List-Post: <mailto:nfsv4@ietf.org>
List-Help: <mailto:nfsv4-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/nfsv4>, <mailto:nfsv4-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 01 Dec 2010 15:48:44 -0000
On 2010-12-01 00:14, Trond Myklebust wrote: > On Tue, 2010-11-30 at 18:18 +0200, Benny Halevy wrote: >> On 2010-11-29 20:53, Trond Myklebust wrote: >>> On Mon, 2010-11-29 at 13:27 -0500, david.noveck@emc.com wrote: >>>> There are three possibilities that occur to me. If there are more, I'd >>>> prefer not to know :-) >>>> >>>> Stateid-that-created the layout: I take your point about usefulness, >>>> and the text doesn't say anything specifically that says this is right. >>>> On the other hand, if it is not used for this (determining when to >>>> return return-on-close segments), what is it used for? It seems like it >>>> should be used for something, if it there. >>>> >>>> Any-close: Not sure this is very useful, but if you were just trying to >>>> read what the spec says, it seems like that is what the spec says. >>>> There are lots of places where you would expect it to say "last close" >>>> if that was what was meant, but I don't see them. Does it say this >>>> anywhere? >>>> >>>> Last-close: This seems the most useful choice but it isn't what the spec >>>> says, so if we can all agree that this is what it should say, maybe the >>>> errata mechanism would be used to fix this as well. >>> >>> Yes, it does appear that a spec clarification is needed here too. >>> >>>> Also, it occurs to me that delegation adds another level >>>> complexity/nightmare and a delegation stateid can be passed just as a >>>> open or lock stateid can. One purpose of the delegation is that >>>> opens/closes on the client, as long as there is a delegation can be >>>> hidden from the server, so I guess this means that the responsibility >>>> for effecting a last-close return, when a delegation is present is for >>>> the client to do by means of LAYOUTRETURN to the server. Given that the >>>> client is in control, he knows the ordering and doesn't have to be >>>> conservative on open/close ordering. >>> >>> Doesn't have to be, but should be allowed to be. >>> >>> I'm gunning for a client position where we don't have to add any extra >>> serialisation between LAYOUTGET and LAYOUTRETURN, but can rely on the >>> forgetful model to let us drop the layout whenever we please. >>> >> >> We need to separate between the layout and the layout stateid. >> The server doesn't care weather the client uses the layout it was given or not. >> That's up to the client. However, the server and client must agree on >> the layout stateid. > > Yes. Why do they not? > One corner issue that I'm concerned about is what stateid results in CB_LAYOUTRECALL that does not cover the whole layout (either byte range or iomode) that returns CB_NOMATCHING_LAYOUT. We want this case to be equivalent to a LAYOUTRETURN for the recalled range, however, unlike LAYOUTRETURN the server does not provide the client with the resulting stateid and if the client dropped the layout segments in the remainder of the recalled range it may believe it has no layout and reset its layout stateid, while the server thinks it still holds a part of the layout and will just update the layout seqid. >>> In such a scenario, the automatic last-close return is actually quite >>> useful, since it allows the server to be forgetful too... >>> >> >> What do you mean by that? > > I mean that the server does not have to wait for and/or recall those > layouts that are close-on-return if the client fails to send > LAYOUTRETURNs before CLOSE. It just forgets them. > OK. I just don't consider this "forgetting" but rather responding to an implicit return... Benny
- [nfsv4] More layout stuff: layout return on close Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Benny Halevy
- Re: [nfsv4] More layout stuff: layout return on c… Tigran Mkrtchyan
- Re: [nfsv4] More layout stuff: layout return on c… Benny Halevy
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Tigran Mkrtchyan
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Tigran Mkrtchyan
- Re: [nfsv4] More layout stuff: layout return on c… david.noveck
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Sorin Faibish
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Sorin Faibish
- Re: [nfsv4] More layout stuff: layout return on c… david.noveck
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… david.noveck
- Re: [nfsv4] More layout stuff: layout return on c… Benny Halevy
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… david.noveck
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… david.noveck
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… david.noveck
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Benny Halevy
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Benny Halevy
- Re: [nfsv4] More layout stuff: layout return on c… Benny Halevy
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Fred Isaman
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… david.noveck
- Re: [nfsv4] More layout stuff: layout return on c… Spencer Shepler
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Benny Halevy
- Re: [nfsv4] More layout stuff: layout return on c… Benny Halevy
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust
- Re: [nfsv4] More layout stuff: layout return on c… Benny Halevy
- Re: [nfsv4] More layout stuff: layout return on c… Trond Myklebust